[
https://issues.apache.org/jira/browse/FLINK-21301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17279710#comment-17279710
]
Andy commented on FLINK-21301:
------------------------------
[~jark] Thanks for your reply. Of course, we could set allow lateness to state
ttl for window state if `lateFireDelayEnabled` is enabled. However, it's more
flexible if we could set allow lateness different with state ttl. For above
example, user may want to set `MinIdleStateRetentionTime` to 1 day, but only
want set allow lateness to 5 min ~10min instead of 1 day.
Looking forward to your reply, thanks.
> Decouple window aggregate allow lateness with state ttl configuration
> ---------------------------------------------------------------------
>
> Key: FLINK-21301
> URL: https://issues.apache.org/jira/browse/FLINK-21301
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Planner
> Reporter: Andy
> Priority: Major
>
> Currently, state retention time config will also effect state clean behavior
> of Window Aggregate, which is unexpected for most users.
> E.g for the following example, User would set `MinIdleStateRetentionTime` to
> 1 Day to clean state in `deduplicate` . However, it will also effects clean
> behavior of window aggregate. For example, 2021-01-04 data would clean at
> 2021-01-06 instead of 2021-01-05.
> {code:sql}
> SELECT
> DATE_FORMAT(tumble_end(ROWTIME ,interval '1' DAY),'yyyy-MM-dd') as stat_time,
> count(1) first_phone_num
> FROM (
> SELECT
> ROWTIME,
> user_id,
> row_number() over(partition by user_id order by ROWTIME ) as rn
> FROM source_kafka_biz_shuidi_sdb_crm_call_record
> ) cal
> where rn =1
> group by tumble(ROWTIME,interval '1' DAY);{code}
> It's better to decouple window aggregate allow lateness with
> `MinIdleStateRetentionTime` .
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
